Contrast-enhanced ultrasound (CEUS) of cystic renal lesions in comparison to CT and MRI in a multicenter setting

2.1 Statistical analysis

Continuous variables were tested for normal distribution using the Kolmogorov-Smirnov-test; variables not following a normal distribution are reported as median and interquartile range (IQR) and were compared using the Mann-Whitney U-Test. Categorical variables were compared using student’s t-test or chi² test, as appropriate. Diagnostic accuracy of CEUS compared to gold standard CT was tested using sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Additionally, exact 95% confidence intervals (CI) were calculated for all values. Intra-class-correlation (ICC) was used to compare Bosniak classes in CEUS to both ceCT and ceMRI. Interpretation of ICC inter-rater agreement was defined by Cicchetti [21]: Less than 0.40 – poor, 0.40 to 0.59 – fair, 0.60 to 0.74 – good and 0.75 to 1.00 as excellent. All statistical analysis was performed using SPSS software version 24. All tests were two-sided and significance defined by a p-value < 0.05.

CEUS scores were equivalent to ceCT in 64/87 (73.6%) with an upgrade of 4 lesions (4.8%) and downgrade of 19 cystic renal lesions (22.9%) by ceCT. (Fig. 1a) Two lesions of Bosniak II were each up- and downgraded by one class and two lesions were downgraded by ceCT from Bosniak IV to III. Representative case of correct scoring in CEUS and ceCT is shown in Fig. 2. More different scorings were found in intermediate complex cysts Bosniak IIF and III: Of 45 Bosniak IIF lesions ceCT rated one lesion as Bosniak I and ten lesions as Bosniak II while three lesions were upgraded into Bosniak III. Five lesions classified as Bosniak III in CEUS were downgraded by ceCT, one as Bosniak II and four as Bosniak IIF. Overall ICC of CEUS and CT was 0.824 (95% -CI: 0.74–0.88; p-value < 0.001). Sensitivity, specificity, PPV and NPV of CEUS compared to gold standard ceCT are presented in Table 2.

OPEN IN VIEWER

Fig.1

Lesion distribution of CEUS and cross-sectional imaging. Bosniak classification in CEUS compared to a) CT and b) MRI. CEUS showed small number of different scoring (21.8%) using Bosniak classification compared to CT (a), while comparison to MRI (b) demonstrated a higher number of mismatches (55.8%). Abbreviations: CEUS denotes contrast-enhanced ultrasound, CT, computed tomography; MRI, magnetic resonance imaging.

OPEN IN VIEWER

Fig.2

Representative case of an excellent correlation of a Bosniak II cystic renal lesion in CEUS and CT. a) Contrast-enhanced CT scan: Subcapsular, hyperdense renal cyst (red circle) without any calcifications. They do not show any solid components or contrast enhancement. https://www.thieme-connect.de/media/ultraschall/201603/uim-1947_10-1055-s-0042-104646-i3.jpg. b) Intraluminal echo in B-mode ultrasound (right side), CEUS (left side) shows no (septal) contrast enhancement of the cyst, therefore the cyst can be classified as Bosniak type II. Abbreviations: CEUS denotes contrast-enhanced ultrasound, CT, computed tomography.

Equivalent scores in CEUS compared to ceMRI were detected in 48/86 cystic lesions (55.8%) with a wide range of both up- and downgrades. (Fig. 1b) No change was detected in Bosniak class I. Two Bosniak II lesions in CEUS were each upgraded (Bosniak IIF) and downgraded (Bosniak I). More downgrades (5/31 into Bosniak I and 7/31 into Bosniak II) and upgrades (One lesion each into Bosniak III and IV) were classified as Bosniak IIF. 12/28 Bosniak III lesions in CEUS were scored different as followed: 1 as Bosniak I, 4 as Bosniak II and 7 as Bosniak IIF with an upgrade of 4 lesions into Bosniak IV. Bosniak class IV showed a downgrade of 2/14 lesions by ceMRI into Bosniak III. ICC of CEUS compared to ceMRI was 0.651 (95% -CI: 0.51–0.76; p < 0.001). Table 3 presents intra-class-correlation of CEUS to both ceCT and ceMRI.

Comparison to histopathological reports was available in 35 cystic lesions which were occurred by total nephrectomy (n = 17/35, 47%), partial nephrectomy (n = 14/35, 39%) and biopsy as part of minimal-invasive therapy e.g. radiofrequency ablation (n = 4/35, 11%). CEUS showed a sensitivity of 88% (95% -CI: 72–95), specificity of 25% (95% -CI: 5–70), a PPV of 90% (95% -CI: 75–94) and a NPV of 20% (95% -CI: 4–62). One cystic lesion with suspicion of malignancy were histologic confirmed as a renal oncocytoma (classified as Bosniak III) and another one as a multilocular cystic neoplasm (classified as Bosniak IV) in histopathological report. Two cystic lesions (Bosniak II and IIF) showed disease progression and morphologic change after two and five years and both were confirmed as renal cell carcinomas.

4.1 Potential of CEUS for the classification of renal cysts compared to CT and MRI

After the introduction and revision of Bosniak’s classification a first modification to MRI was described by the same research group in 2004 [22]. Our findings agree with the results of single-center studies who compared the accuracy of CEUS with ceCT in a small number of cases. With an equivalent classification of cystic renal lesions in 73.6% compared to ceCT presenting similar results compared to smaller studies by Park et al. 74% and a slightly lower percentage than Clevert et al. 81% [17, 19]. Prospective single-center studies presented comparable results with a correct rating in 79% of all cystic renal lesions (n = 46) similar to the classification by ceMRI in that study [23]. CEUS showed an upgrade in 19/87 cystic lesion (21.8%) compared to gold standard CT especially in intermediate complex lesions Bosniak IIF (n = 13/45) and Bosniak III (n = 5/10). (Fig. 2) ICC in CEUS and ceCT showed an excellent correlation only in Bosniak IV group while all other groups presented a good (Bosniak I) or moderate correlation (Bosniak II-III). ICC in CEUS and ceMRI was lower in each Bosniak group compared to ceCT which remains the standard imaging tool for Bosniak classification of cystic renal lesions (Table 3). Due to a high spatial resolution US improves the detection of finest septae and nodular wall thickening. Since CEUS is able to detect very small amounts of microbubbles, it has the power to visualize even finest septal enhancement, which is often missed in cross sectional imaging. (Fig. 3)

OPEN IN VIEWER

Fig.3

Different classification of a cystic renal lesion in CT (Bosniak II) compared to CEUS (Bosniak IIF) due to the sensitive visualization of septal contrast enhancement. A,b) Axial and coronal contrast-enhanced CT scan: a cystic renal lesion of the right kidney without wall thickening or calcifications. Detection of a hairline thin septa without contrast enhancement. This cyst would be classified as Bosniak type II. c) B-mode ultrasound: Detection of more intraluminal septa with nodular thickening due to a high spatial resolution. d) No vascularization of the thin septa using colour-coded Doppler ultrasound. e) Visualization of microvascularization using innovative ultrasound Doppler technique superb micro-vascular imaging (SMI). f) Contrast-enhanced ultrasound of the cyst shows nodular contrast enhancement of the hairline thin septa. Therefore this cyst can be classified as Bosniak type IIF. Abbreviations: CT denotes computed tomography; CEUS, contrast-enhanced ultrasound.

Diagnostic work-up and stratification of these intermediate complex cystic lesions are important to differentiate surgical procedures or follow-up of these lesions [24]. While a different rating in Bosniak II and IIF lesions only results in a change of follow-up time interval, surgical removal of Bosniak III lesions is recommended [25]. The sensitive detection of microbubbles in CEUS with the visualization of fine septal enhancement and small nodular wall enhancement as a potential to improve the classification of cysts and may result in an upgrade of lesions into a higher Bosniak class [13, 27]. Contrast material–specific software allows subtraction of background tissue from the image so that minimal amounts of US contrast enhancement can be visualize which may not be identified at B-Mode US, ceCT or ceMRI [28]. (Fig. 4) While number and thickness (>1 mm) of septa and its fine contrast enhancement are major criteria for Bosniak classification [26], a high spatial resolution in ultrasound may even visualize more septa and irregular wall thickness. CEUS was superior in detecting the degree of septal wall thickening, septal enhancement, and enhancement of solid components within the lesion compared to CT [19]. Recent studies have demonstrated that the accuracy of intravenous ultrasound contrast agent for detection of septal enhancement in complex cystic lesions was superior to CT [17–19, 29].

OPEN IN VIEWER

Fig.4

CEUS software allows subtraction of background tissue from the image to visualize minimal amounts of septal contrast enhancement. a) B-Mode US using a mid-class ultrasound system: Limited interpretation of the cystic lesion (white arrow) with focal calcification and intraluminal echoes. b) B-mode US using a high-class ultrasound system: Higher resolution but still insufficient interpretation of the renal cyst with intraluminal echo due to focal calcification. c) Visualization of minimal septal contrast enhancement using CEUS. Abbreviations: CEUS denotes contrast-enhanced ultrasound.

Ultrasound and especially CEUS are fast and cost-effective clinical methods of choice for further characterization of unclear cystic renal lesions in CT or MRI [15, 30]. Nevertheless, while the amount of contrast agent depends on the US system, an examination of both kidneys and the use of two contrast agent boli can lead to higher costs and longer examination time by the need of a delay up to ten minutes for contrast agent washout prior using a second bolus injection. The most important differential diagnosis of unclear cystic renal lesions is the (cystic) renal cell carcinoma. Larger renal cell carcinoma shows area of necrosis or haemorrhage that can be detected and misinterpreted as cystic lesions in B-mode ultrasound [5]. Evaluation of these solid (enhancing) components can be evaluated by the use of contrast media and a dynamic examination. CEUS is an useful method to differentiate between malignant and benign (cystic) renal lesions with a high diagnostic sensitivity according large single-center studies using CEUS with a histopathological validation [31]. Although the specificity in our subgroup analysis remains low, there is only a very small number of benign lesions (n = 4/35) in this retrospective cohort. Since the Bosniak classification has worked well using ceCT, it may be appropriate to modify it as opposed to the development of a separate classification for CEUS. If a scoring system for contrast enhancement positive and negative lesions on CEUS is used (regardless of the Bosniak classification), the NPV based on the reference was 100% (0/842 lesions), whereas the PPV was 97% (170/175 lesions) [28].

Different advantages and disadvantages of each modality (CT, MRI, CEUS) in visualization and interpretation can result in a mismatch of Bosniak classification. New cross-sectional imaging methods like multiphase dual energy CT with material density iodine datasets can increase readers’ confidence for renal lesion detection and characterization [32]. Further-more, a meta-analysis of 764 patients by Lassel and colleagues showed that diffusion-weighted MRI and apparent diffusion coefficient (ADC) values can help distinguish between benign and ma-lignanttumors, potentially providing a more efficient radiologist workflow and reducing the number of unnecessarily performed interventions [33].

Using CEUS, patient compliance is always required and image interpretation in patients with obesity or bowel gas is still limited due to the physics of ultrasound. Besides operator-depending limitations, CEUS is a good diagnostic alternative for patients with impaired renal function without side-effects and radiation exposure [15, 34]. Since the effect of iodinated contrast media in renal insufficiency is well known and recently published studies describes a consistent influence on the kidney perfusion after repeated injections [35], we actually do not know the long term effect of gadolinium retention in brain after many injections during follow up [36]. CEUS has advantages over contrast-enhanced CT and MRI including unmatched temporal resolution due to continuous real-time imaging [27]. Based on a retrospective analysis by Barr et al. including more than 1018 renal masses CEUS can provide increased confidence that Bosniak II and IIF lesions are benign [28, 37]. With a high PPV and NPV of CEUS, the evaluation of cystic renal masses could decrease the need for CT, with its associated radiation, or MR imaging, with its higher costs. CEUS was cost-efficient compared to both ceCT and ceMRI in the characterization of unclear liver lesion [27, 38]. The use of US and especially CEUS as an additional examination method in a multimodal imaging algorithm has beneficial value in both detection and characterization of small lesions (<10 mm), while CEUS can even visualize thin septae and septal enhancement but has limitations in deeper lesions or due to obesity or bowel gas. While complex cystic renal lesions in polycystic kidneys are often an incidental finding on B-Mode US or cross-sectional imaging, combination of CEUS and ceCT/ceMRI by the use of newer image fusion techniques may increase the diagnostic performance can improve identifiability and characterization of renal tumors in a multimodal approach.

While CEUS needs a high experience for both examination and image interpretation, the need of a nationwide education in CEUS imaging leaded by experts in the field of ultrasound and CEUS is indispensable to guarantee high quality standards.

4.4 Limitations

Main limitation is the small number of histopathological reports since CEUS is compared to either ceCT or ceMRI. Patients with lesions assessed by ceMRI had no additional ceCT for direct comparison. CeMRI is however regarded to be an accurate imaging tool for the assessment of cystic lesions during the last decade and widely used across imaging of cystic renal lesions avoiding radiation exposure. Diagnostic assessment and classification of the renal cysts in ceCT and ceMRI were performed by different local readers. Due to the retrospective nature of this study a more detailed presentation of the reasons for up- and downgrading is difficult. Prospective studies are necessary to evaluate exact reasons for different rating of cystic renal lesions in CEUS compared to other modalities especially gold standard CT.

4.5 Conclusion

This study demonstrated that the Bosniak classification in CEUS showed an excellent correlation compared to gold standard imaging ceCT. A major caveat is that applying Bosniak criteria, CEUS upgrades a percentage of lesions compared to ceCT, potential driven by the increased sensitivity of CEUS for septal enhancement. While an upgrade in CEUS might change the workup and management of cystic renal lesions, a modification of the Bosniak classification with special emphasis on septal contrast enhancement pattern in intermediate complex cystic lesions may reduce unnecessary biopsies and surgery by different scoring in CEUS as well as selecting patients for surgery who would otherwise be monitored.